Computing the Elusive Spirit of Place by Inderjeet Mani

We have our entanglements and love affairs with places. “And the end of all our exploring,” T. S. Eliot promises,will be to arrive where we started and know the place for the first time. Such knowledge may be never-ending.

We don’t know what the sense of place felt like to our hunter-gatherer forebears, but judging from their sophisticated tracking and navigational skills, they were able to notice things in their environment that most of us have long forgotten. Luckily, along with other mammals, we still have our built-in sense of place, with maps of places we visit stored at a neural level within our brains. The particular environments we move in activate ‘place cells’ in our hippocampus, and these cells also help us remember past experiences of the spatial environment. Together with other cells in the nearby entorhinal cortex, they create an inner GPS that makes it possible for us to navigate to our desired destinations. This remarkable evolutionary apparatus forms the underpinning for our higher-level sense of place.

Today, our technologies provide complex visualizations of places and journeys through them long before we get there. Advances in telescopy and remote sensing are opening up worlds of which we had only a faint inkling. Systems like Google Earth efficiently integrate billions of bits of data from satellite and aerial imagery, street photographs, and consumer-relevant information, serving up maps at different scales and navigational directions to us, all tailored to our location and anything else the system knows or guesses about us. We use them everywhere even though our map-based systems flagrantly distort the earth’s shapes, areas, distances, or directions (inherent in the problem of projecting our ellipsoidal blob onto a 2D display), and we only occasionally wake up to the fact that maps can encode prejudices. For our mapping systems are undoubtedly useful, even helping people find their birthplace and locate the dead. Do these and the even more advanced technological capabilities described below enhance our sense of place, or compromise it?

Luring us in person or on the wings of memory, a place gives us the chance to luxuriate in its setting, and also to touch and be touched by other pasts. I will never forget my first sight of the Roman town of Volubilis, nestling in a gentle Moroccan valley amid olive groves with the blue wash of the Atlas Mountains rising in the distance. Disembarking from my long bus ride, I hurried over to the ruins. I was strolling across the dusty floor of a ruined villa when something made me stop in my tracks. Under my feet lay a circular mosaic, partly chipped away, revealing a bewitching Orpheus strumming his lyre to a circle of crudely-drawn African animals. The present and past seemed to touch each other in a circle, with the dusty and ephemeral imprint of my hiking boot a link along the way.

Modern though we may feel, the atmospherics of a place can affect us in ways that might have been familiar to our ancestors. I was visiting Angkor Wat in Cambodia when I took off for a hike in nearby Kbal Spean along the River of a Thousand Lingas. There, shimmering on the riverbed lay phallic Shivas and reclining Vishnus, their splendid carved figures washed by those mineral-rich waters for nearly a millennium. The sight of those old gods for some reason made my atheist spirits soar, and with that came a renewed sense of unbelonging and freedom. I felt I was entering a liminal zone, a border between the rational and the mystical, and as if to amplify such magical thinking, which may have been of ancestral origin, the breeze dropped off and the current seemed to rush forward with a sudden burst of sound. Our Cambodian guide stopped, lit a cigarette, and explained that the atmosphere had been charged over centuries by the spiritual exertions of hermits in the cliffside caves along the way.

Thinking of cliffs, particularly those free from the influences of hermits, I am reminded of the walls of Californian sandstone at Drakes Bay where I spent many an afternoon ambling past slumbering elephant seals, admiring the soft sheen of the flanks of these visitors who have been coming to these beaches since ancient times to mate and wean their young. In the sparkling air cooled by a green Pacific, it was hard not to feel one in the moment with the seals as well as the flitting swarms of tiny seabirds who (one would like to think) were sharing the bracing sense of being alive. And yet in my walks on those sands, my thoughts often turned to another visitor from the past, to what the explorer and privateer Sir Francis Drake must have felt in 1579 when his footprints more or less overlapped with mine after coming ashore there during his three-year circumnavigation of the globe. Today, as location information accumulates deeper within computer systems, it becomes possible to generate (with the right privacy protections) a visitor log for any place, giving us a feel for the chain of footsteps in which we follow, including those of other explorers and writers.

Writers have always waxed eloquent about the sense of place. In my youth, I was drawn to D. H. Lawrence, and as soon as I could afford it I followed his footsteps to Taos (where visitors can save a pine cone from his favorite tree) and Tuscany (where they can view astonishingly beautiful Etruscan sarcophagi at the Guarnacci Museum in Volterra). Lawrence’s essay The Spirit of Place includes the following mystical passage:

Different places on the face of the earth have different vital effluence, different vibration, different chemical exhalation, different polarity with different stars: call it what you like. But the spirit of place is a great reality.

Our sense of place is a great reality indeed, but subjective and mysterious in most formulations. Even those charged with heritage preservation have a hard time wrapping their arms around it. The International Council on Monuments and Sites settled on the following definition, called the Quebec Declaration:

Spirit of place is defined as the tangible (buildings, sites, landscapes, routes, objects) and the intangible elements (memories, narratives, written documents, rituals, festivals, traditional knowledge, values, textures, colors, odors, etc.), that is to say the physical and the spiritual elements that give meaning, value, emotion and mystery to place.

Tangling with intangibles (even if some appear tangible) seems hopeless, and one has to look to geography and environmental planning to get some traction on the concept. In his book Space and Place, the geographer Yi-Fu Tuan distinguishes three facets of our sense of place: the traditional spirit of place, or its genius loci, literally the nature spirits who inhabit profane space and give it a sacred and awe-inspiring quality; the personality of place, where a place is seen as acquiring “unique signatures in the course of time…through the prolonged interaction between nature and man”, inspiring awe or affection; and the sense of place proper, which has to do with a place’s aesthetic appeal, via the eye as well as the other senses including touch and odor, and involving a more profound and meaningful knowing of and attachment to a place that we become aware of “only when we have left it and can see it as a whole from a distance”. My complex experiences at Volubilis, Kbal Spean, and Drakes Bay may be understood phenomenologically in terms of the places’ unique personality and my interactively-derived sense of place, with Kbal Spean in addition having a longstanding relation with nature spirits. I remain attracted to these spots, especially while seeing them from a distance today through my mind’s eye and via online visualizations.

Phenomenology is an interpretive discipline, not one that lends itself to formalization, least of all for use by computers. However, landscape planners have long cherished the idea that the spirit of place might be evoked through design and instrumentation. Some of them may follow the 18th-Century advice of Alexander Pope:

Consult the genius of the place in all;
That tells the waters or to rise, or fall;
Or helps th’ ambitious hill the heav’ns to scale,
Or scoops in circling theatres the vale;
Calls in the country, catches opening glades,
Joins willing woods, and varies shades from shades,
Now breaks, or now directs, th’ intending lines;
Paints as you plant, and, as you work, designs.

A study by environmental scientists at Australia’s Commonwealth Scientific and Industrial Research Organization (CSIRO) and Penn State has further operationalized our sense of place. They view it in terms of three psychological dimensions: the cognitive dimension has to with our beliefs about a place’s role in defining who we are, the affective with our emotional attachment to a place, and the conative with its behavioral advantages, i.e., how the place allows us to interact with it in satisfying ways. Surveying the sense of place among property owners around eight lakes in Wisconsin, the results showed that that our sense of place is not only far from random, but predictable. The most consistent predictor of the sense of place was the subjects’ attitude to nature (in this case, towards preserving native vegetation). People who felt that the lake was important also believed that it helped define who they were, suggesting that our relationship to nature is tied to beliefs about the self (and personal freedom).

Given the predictability of the sense of place based on this sample, it is an interesting question as to whether computing might predict it for users in the large. Prior research by Yahoo Labs in collaboration with the University of Cambridge is encouraging. Using crowdsourced judgments of pictures, the system homes in on urban areas in major cities that are judged to be more happy, beautiful, or peaceful. In London, for example, the happiest spots are those positively correlated with features such as greenery, open space, and presence of red-brick homes, all of which features are automatically detected in the pictures using computer vision.

Computing the spirit of a place can take advantage of our ability to experience the unique historical signature mentioned by Yi-Fu Tuan, drawing us to a place’s past. When revisiting London recently after living there long ago, I found it useful to rewind Google’s street views back in time to my old neighborhood. My walks in the city were further enhanced by a sequence of maps developed at the Center for Advanced Spatial Analysis at University College, London showing how today’s vast metropolis evolved from a first-century Roman river garrison. And when planning a visit to Rome, I was drawn to an inspired use of 3D digital models, from the Virtual World Heritage Lab at the University of Virginia. The Lab has created imagined cityscapes consistent with the historical record, yielding this fly-through of Rome in the year 320, where a densely built-up and artificially neat city awaits the viewer. Such armchair exploration can also help those who are physically challenged or otherwise unable to visit.

User feedback about the spirit of place need not be confined to crowdsourced information. A major source of inspiration about place atmospherics is travel narratives. Computers are now able to process these texts, geocoding place references where possible to auto-construct maps of the perambulations of characters. Research I carried out with teams at Georgetown, MITRE, and Brandeis University has developed systems that could automatically glean travelers’ paths in space and time from contemporary sources such as online biking blogs. Scientists at the Universities of Pau and Bordeaux have also been able to carry out similar computations from a collection of 19th-Century French travelogues involving journeys in the Pyrenees mountains, and (as described in another essay) folklorists at UCLA have in turn been able to use historical gazetteers to auto-generate maps of significant entities and events in collections of historical Danish folktales, such as this demo of witchcraft-related information. Bearing in mind that maps with paths marked on them are like narratives in themselves, these visualizations allow a visitor to explore summaries of what different travelers have noted along the way. They also provide features for automatically learning users’ place-related interests.

Travel websites have long been able to recommend places to tourists that are similar to what they have preferred in the past and to similar places that tourists with similar tastes (and similar tourism-related ratings for places) have preferred. When such recommender systems are naturally extended to accommodate spirit-of-place ratings, they can leverage additional user information (including environmental attitudes and descriptive terms mined from geocoded travel narratives), along with place features that are geographic (such as location, land forms, vegetation, terrain, climate, air quality, open space) as well as crucial pictorial information.

For visual features of places, computer vision systems are now able to learn from datasets like the Places database from MIT, which has 10 million color photographs of scenes labeled with categories and attributes. Using their neural network system demo, the above panorama of Volubilis (which was new to the system) generated three scene categories that the net was somewhat diffident about: ruin, valley, and butte, with the scene attributes indicating it was an open area, bathed in natural light, sunny, with vegetation, a rugged scene, with a far-away horizon and opportunities for hiking, with a man-made structure but also natural. These cognitive and conative attributes definitely contributed to my experience of the sense of place. The MIT system was far more confident (at 84%) about the Orpheus mosaic (also new to the system), classifying the scene as an amphitheater, inferring that based on the ledge on the right. The accuracy of such systems is likely to improve much more in the near future, given the number of competitive tasks being organized to address automatic scene classification, along with the rapid pace of advances in neural net technology. Overall, this makes me confident that the spirit of place engines are coming.

When I first traveled to the places mentioned here, it was in the previous century. I had no sophisticated tools at hand. I remember arriving at Drakes Bay in a beat-up rental car that had stalled more than once. I consulted a paper map in advance and wrote up my own driving directions, and as I drove, I regularly checked road signs and landmarks to make my way. Travelers today arrive at the beach well-prepared for what lies ahead, guided by GPS and blindly following their machines’ spoken directions. Tomorrow’s wayfinders will consult maps far less, lacking perhaps any knowledge of the path they are taking while being ferried to that beach by self-driving vehicles and Jetsons-like drones. As with arithmetic calculation, why bother, when the machine will figure it out for you?

GPS is often indispensable, but it comes with costs. A landmark study by spatial cognition researchers at the University of Tokyo compared walkers using GPS with those using paper maps and others who had direct experience of routes, finding among other things that the GPS-users drew less accurate sketch maps. As shown by subsequent research at the Environmental Systems Research Institute (ESRI) in California, whether people use GPS or paper maps, they don’t fare well on measures of environmental spatial ability, which includes their facility with interpreting maps, giving directions, judging distance, remembering routes, and so forth. In other words, we are becoming the sort of people you don’t want to rely on for directions while on a road trip, let alone an expedition. No wonder there are increasing news reports of smart-phone brandishing city slickers having to be rescued from the wilderness. Such spatial illiteracy may impair our memory of place, a key aspect of the sense of place. We may take pictures as well as mental snapshots of the scenery to write about later, but not having to grapple with navigation will likely make us store in our hippocampus fewer or less-detailed maps of the places we visit.

Higher spatial literacy may have been more common in ancient times, when a person had to navigate in their environment relying on the heavens, winds, weather, ocean currents, and landmarks, as well as the paths of other moving creatures. Today, vestiges of such spatial literacy continue to flourish in certain languages. As shown in field experiments by scientists from the Max Planck Institute for Psycholinguistics in the Netherlands, speakers today of the Australian language Guugu Yimithirr routinely say things like “that ant is just north of my foot”, knowing their correct position and orientation no matter where they are and what the weather is like. Likewise, the nearly half a million native speakers of Tzeltal, spoken in Chiapas, Mexico, say things like “the bottle is uphill of the chair”, and correctly identify their position and orientation even when spun around blindfolded in a darkened house. Similar examples exist for languages spoken in India (Kota), Nepal (Belhare), and elsewhere. The expressiveness of these and other spatial expressions can also be represented in AI systems, as shown in a recent book I co-authored.

We moderns can still improve our spatial literacy. This is perhaps best exemplified by the cabbies of London, who are certified based on their ability to get to 25,000 streets and 100,000 landmarks. Research at University College, London has shown that they have swellings in their hippocampus, and these swellings subside upon retirement. However, the price of progress is the dumbing down of such capabilities: these spatially super-literate taxi drivers are now in a life-and-death struggle with Uber, though there may still be hope for them.

The increasing virtualization of our environments is also likely to interfere with our sense of place. The fact that we are gazing at screens so much while traveling, and in future walking while seeing the world through VR headsets or smart eyewear, is likely to make our sense of place an ersatz hybrid of the natural and the artificial. To have a display overlaid on part of your visual field, with the rest devoted to seeing the natural world is distracting enough, let alone the temptation to distract yourself even further with the latest messages or taking pictures. When you don’t pay attention due to multitasking, you are likely to observe and remember a place less well. In a recent psychological experiment from Fairfield University in Connecticut, people who were asked to photograph museum exhibits had trouble recalling them, while exhibits they didn’t photograph proved easier to remember.

Of course, as virtually-fixated spatial illiterates, we may have plenty of time to gaze out of our smart car or drone windows and enjoy the scenery en route, even when we view it through the lens of an artificial device. Our adaptation to these technologies may not fundamentally damage our biological ability to perceive a place. But our inability to give others directions may also leave us vulnerable when device-free, and unable to converse fluently about where the amazing place we have just ‘discovered’ actually is.

Our absorption in hybrid, digitally-mediated views of places is also likely to eliminate the serendipity of uncovering aspects of places that endear us, while also perhaps exposing us to danger. A future traveler to Volubilis may miss the Thracian musician in the dust beneath his feet unless his device directs him to it. When hiking in Kbal Spean, GPS reception is poor, and the traveler watching the old gods had better not lose sight of his guide. And on Drakes Bay, he will need to keep an eye out for the bull seals, who do not take lightly to posing for selfies. By then, our own authentic sense of any of these places may have been deluged in advance by suggested computer-supplied information spun to us through engaging visualizations. However, while powerful collective influences shape our expectations today, culture has always held its sway. I was certainly biased by reading Lawrence before visiting those Etruscan tombs, and many a pilgrim may be in a state of exaltation in advance of visiting a cherished shrine. These expectations may stimulate our travel to a place, but they may also impede us in our journey towards knowing the place for the first time.

Inderjeet Mani
Inderjeet Mani is a writer and scientist based in Thailand. His books include a study on time called The Imagined Moment. To read more of his published essays and stories, please his site.

Follow us!
Share this post with your friends.

Leave a Reply

Your email address will not be published.